Predicting species response to climate change is a central challenge in ecology, particularly for species that inhabit large geographic areas. The mountain pine beetle (MPB) is a significant tree mortality agent in western North America with a distribution limited by climate. Recent warming has caused large-scale MPB population outbreaks within its historical distribution, in addition to migration northward in western Canada. The relative roles of genetic and environmental sources of variation governing MPB capacity to persist-in-place in a changing climate, and the migratory potential at its southern range edge in the United States, have not been investigated. We reciprocally translocated MPB populations taken from the core and southern edge of their range, and simultaneously translocated both populations to a warmer, low-elevation site near the southern range boundary where MPB activity has historically been absent despite suitable hosts. We found genetic variability and extensive plasticity in multiple fitness traits that would allow both populations to persist in a warming climate that resembles the thermal regime of our low-elevation site. We demonstrate, for the first time, that supercooling points in MPBs are influenced both by genetic and environmental factors. Both populations reproduced with seasonally appropriate univoltine generation times at all translocated sites, and bivoltinism was not observed. The highest reproductive success occurred at the warmest, out-of-range low-elevation site, suggesting that southward migration may not be temperature-limited.

MPB collection, tree harvest, and bolt infestation

For both field and laboratory experiments, we felled MPB-infested trees on 21 June 2016 from the S_UT site and on 4 May 2016 from the S_AZ-high site (Table 1). Cut bolts (~46 cm long) were harvested from one tree at each of the S_UT and S_AZ-high sites. Bolt ends were sealed with paraffin wax and transported to the Rocky Mountain Research Station (RMRS) Laboratory in Logan, UT where they were placed at ambient room temperature to allow adults to emerge naturally. Adult beetles were collected daily and stored in Petri dishes lined with distilled water-moistened filter paper at 4°C for up to approximately 10 days. To rear the next generation of beetles we also harvested three live, healthy trees of the same species at each site, cut them into ~46 cm long bolts, and sealed the cut ends with paraffin wax to retain moisture and deter fungal contamination. Bolts were stored at 4°C for up to 3 weeks. The uninfested experimental bolts from each site were randomized among the three field sites and the two temperatures in the laboratory study.

We determined the sex of emerged adult beetles using the morphologically distinct 7^th tergite (Lyon 1958). To avoid potential genetic differences in development time among emerging adults, and to standardize for cohort density, we used beetles that emerged during the time beginning just before and throughout peak emergence from natal bolts. Experimental bolts of the same species were infested by drilling a small hole vertically into the phloem at the anatomical bottom of the bolt, inserting first a female then a male beetle, and stapling a mesh screen over each hole to prevent beetle escape. To minimize potential maternal effects due to host species (Burke and Carroll 2017), P_UT were reared in P. flexilis and P_AZ in P. flexilis/ strobiformis hybrids that were harvested from the same locations as infested bolts (Table 1). Individuals were randomized by sex and mating pairs were infested 6 cm apart, with 10 to 13 pairs per bolt depending on bolt circumference. Following infestation, bolts were inverted to allow for natural upward gallery excavation. Infested bolts were either transported to field sites or placed in laboratory incubators as described below.

Field Experiments

Experiment set-up

For the field reciprocal translocation experiment, we enclosed infested bolts individually in escape-proof netting (Rothco, MPN 8088) and within 24 hours of infestation, suspended each bolt ~ one meter above the ground in wooden A-frame structures with metal covers at each location (Table 1; Appendix S1: Fig. S1). Prior to infestation bolts were randomly distributed to their respective treatment groupings. Nine bolts infested with the P_UT population and nine bolts infested with the P_AZ population (18 bolts total per site) were placed at each of the three sites (54 bolts total). Bolt location was randomized among three A-frame structures at each site such that there was an equal number of P_UT and P_AZ infested bolts per A-frame (6 bolts per structure) (Appendix S1: Fig. S1). Field experiments were initiated as follows: S_UT: on 30 July 2016; S_AZ-high: on 10 August 2016; S_AZ-low: on 11 August 2016.

To capture thermal conditions at each field site, temperature probes were inserted into the phloem on the south aspect of each infested bolt and temperatures were recorded hourly over the duration of MPB development and emergence (i.e., August 2016 to August 2017) (CR1000, Campbell Scientific, Logan, UT). To describe environmental effects, growing degree hours (GDH) >10°C and <0°C (i.e., cumulative heat and cold units) and weekly maximum and minimum temperatures were calculated for each bolt beginning on 12 August 2016.

MPB collection and trait measurements

Adult brood emergence at each site was monitored at least twice weekly throughout emergence and daily during the weeks of peak emergence. We collected individuals by bolt and transported them on ice to either Northern Arizona University, Flagstaff, AZ, or the RMRS Laboratory in Logan, UT. Generation time for each individual was calculated as the time difference between bolt infestation and brood adult emergence. Generation time includes the time duration for mating, oviposition, egg, larval, pupal and teneral adult (pre-emergent) development, in addition to a facultative prepupal diapause. We considered generation time resulting in seasonally appropriate (i.e., summer) adult emergence to represent higher relative fitness than a generation time resulting in aseasonal emergence. Adult emergence synchrony is important to successful mass attacks and colonization of new host trees (Logan and Bentz 1999). We define emergence synchrony as the standard deviation in generation time across all individuals of a population at a site, where a lower standard deviation suggests greater emergence synchrony and therefore greater fitness (see section Statistical Analyses). Reproductive success, a direct measure of fitness representing number of offspring produced, was calculated as the number of emerged brood adults per bolt divided by the number of successful galleries within the bolt, thereby compensating for uneven mating success and subsequent brood production among bolts. A parent gallery was considered successful (and therefore included in the count) if the gallery length was greater than 10 cm (Eidson et al. 2018), assuming that galleries less than 10 cm were the result of failed copulation by the inserted mating pair. The subset of bolts sampled for cold-hardening (see below) were not included in the determination of reproductive success, as the removal of larvae altered the number of emerged brood.

To measure larval cold-hardening, individual larvae were collected from three infested bolts per population at each field site three times throughout the annual generation: (1) late November/early December 2016, (2) late January/early February 2017, and (3) late March/early April 2017. To account for temperature variability due to bolt aspect, we randomly sampled MPB larvae on three aspects (N, SW, and SE) along the bolt circumference, with each population at each site sampled from all three aspects (one aspect per bolt) each sampling period. To extract larvae, the outer bark and phloem were removed using a 15 cm hole-saw, and the wound was sealed with paraffin wax. Larvae were placed in Petri dishes with filter paper and transported directly or overnight-shipped on ice to the RMRS Laboratory in Logan, UT. Larval instar was determined based on head capsule width (P_UT: Logan et al. 1998; P_AZ: Bentz unpublished).

Supercooling points (i.e., the temperature of hemolymph crystallization) (Lee 1989) of larvae were analyzed within 24 hours of collection. Supercooling points of collected larvae were measured following the protocol of Bentz and Mullins (1999). Briefly, the temperature of individual larvae was monitored while the environmental temperature was lowered at a rate of ~1.5°C min^-1. The supercooling point of each larvae was estimated as the lowest recorded temperature prior to tissue freezing, which was observed as an increase in temperature (≥ 0.5°C) caused by the exothermic latent heat of crystallization. MPB typically has four larval instars prior to pupation, and we observed some combination of larval instars 2, 3, and 4 in cold-hardening samples taken in the fall, winter and spring at each site. No other life stages were observed during sampling for this study. To assess population source and environmental differences in life stage development, we calculated a ‘developmental index’ by averaging instar number (i.e., instars 2, 3, 4) across all observed individuals at each field site and seasonal sampling period.

Adult pronotal width was measured and sex determined (Lyon 1958) for 6,251 individuals (65% of total emerged brood adults). Individuals were collected for sex determination and size measurement at least every 4 days, and every 2 days during peak emergence. We measured pronotal width as a proxy for size (Kozol et al. 1988) using an ocular micrometer to the nearest 0.01 mm.

Laboratory Experiments

We reared each population in laboratory incubators at a constant 18°C and 25°C with a 12:12 hr. photoperiod (Appendix S1: Fig. S2). Optimal larval development in the laboratory occurs at ~25°C for P_UT (Régnière et al. 2012) and ~27°C for P_AZ (McManis et al. 2018). 18°C was used because it is the lowest temperature, in either population, where the majority of individuals can develop directly to the adult stage without induction of a facultative prepupal diapause (Bentz and Hansen 2017). Induction of the prepupal diapause would delay development, and because the two populations differ in diapause intensity (Bentz and Hansen 2017), the developmental delay would generate confounding differences between the populations. Four infested bolts of each population were reared at 18°C and three bolts for each population were reared at 25°C. Adult emergence from individual bolts was monitored daily. Generation time for each individual brood adult and reproductive success per bolt were measured as described in the section Field Experiments. Sex and pronotal size (mm) of 1,532 individuals (31% of total emerged brood adults) were measured as described above, collected from a weekly random population subsample.

Contents of data files are briefly described below. Each data file has a 'metadata' tab with column header descriptions

dryad_data_field_temp.xlsx - phloem temperature data (south aspect) from field sites

• dryad_data_field.xlsx - mountain pine beetle data from field experiment, excluding supercooling and offspring success data (see below)
• dryad_data_field_supercooling.xlsx - mountain pine beetle data from field experiment, supercooling data
• dryad_data_field_offspringsuccess.xlsx - mountain pine beetle data from field experiment, offspring success data
• dryad_data_laboratory.xlsx - mountain pine beetle data from laboratory experiment, excluding size and sex data
• dryad_data_laboratory_size_sex.xlsx - mountain pine beetle data from laboratory experiment, size and sex data